EESemi.com FAQ

- questions most frequently asked by visitors of www.EESemi.com

Electrical Test Parameter Cpk

What does the Cpk number for an electrical test parameter stand for?

The calculated Cpk for an electrical test parameter has the same meaning as the Cpk values calculated for the more commonly-encountered measured quantities that are physical in nature.  For a definition and brief explanation of what Cpk is, please click here.

Cpk is basically a measure of how well the output of your process meets its lower and upper specifications. It is only applicable when the output response exhibits a normal distribution.  Cpk measures two things: 1) how close the mean of the readings are to the center of the lower and upper spec limits (ideally, the mean of the readings must equal the center of the spec limits); and 2) how widely spread the readings are (ideally, the standard deviation of the readings should be zero) .  The higher the Cpk, the better is the capability of the process to meet its requirements.

In the context of electrical parameter testing, the same Cpk calculation concepts apply. Thus, Cpk may only be computed in test parameters that have actual numerical measurements that form a normal distribution, and the measurement readings should come from a statistically significant number of sample units belonging to a homogeneous population.

Consider a test parameter that meets the requirements for Cpk computation.  A certain number of units are tested for this parameter, and all the measurement data are then collected.  The mean and standard deviation of the data are then used to compute for the Cpk of that test parameter, using the same Cpk equation shown in the link above.

As mentioned, the higher the computed Cpk is, the better.  In the semiconductor industry, a Cpk of less than 1.66 needs a closer look.  A Cpk that's less than 1.33 needs some action to make it higher, and a Cpk of less than 1.0 means that the process is not capable of meeting its requirements, and therefore requires an even more urgent response.

The lower the Cpk of an electrical test parameter, the higher is the probability that a device from the same process will fail this parameter.  When analyzing a Cpk, it is important to determine if: 1) the mean of the readings are close to or far from the center of the specs; and 2) the standard deviation of the readings are low or high.  A low Cpk means one of 3 things: the mean is far from the center of the specs, or the standard deviation of the readings is high (i.e., the readings are widely spread), or both conditions exist.

The corrective actions needed to move the readings closer to the center of the specs are generally different from the corrective actions needed to reduce the standard deviation of the readings. In any case, the test engineer must understand why the readings behave in a certain way, and define the corrective actions needed to rectify the situation.

Possible corrective actions include but are not limited to:  exclusion of invalid data that are proven to be due to unrelated assignable causes, repair or substitution of defective test hardware or software used in the test, redesign of the test hardware and software to make it more stable and accurate; redefinition of the lower and upper spec limits of the test parameter based on the actual capability of the device and test system; or even redesign of the device circuit and packaging itself.  Note that it may be necessary to collect a new set of data to recalculate the Cpk of a test parameter if the data initially collected are proven to be a misrepresentation of the actual process capability.